3.5 Microbial composition turnover among regions and host plants
Microbiome composition comparisons between the putative native South Africa and introduced range of medfly were only statistically significant for the pairwise comparisons with Spain (F  = 6.4, R2  = 0.30, p  = 0.015), Colombia (F  = 7.4, R2  = 0.26, p  = 0.015) and Brazil ( = 7.8, R2  = 0.41, p  = 0.015). In the introduced range, the microbiome structure of Spain was significantly different to that of Israel (F =  3.6, R2 =  0.15, p =  0.015) despite being collected partly on the same host plant (Fig. 1) and to Brazil (F =  6.3, R2 =  0.28, p =  0.015). Conversely, the microbial communities from Spain, Australia and Colombia showed no significant differences despite their geographic distances and being collected from different host plants. In the neotropical region, the microbiome of medflies from Brazil and Colombia showed significant differences (F =  7.0, R2 =  0.24, p =  0.015), although collected from different host plants (Fig. 1). ANOSIM analysis at the global scale showed significant differences in samples collected from different locations (R = 0.429, p = 0.001) and different host plants (R = 0.196, p = 0.001), demonstrating the greater importance of geographical distance for microbiome composition compared to diet.
Exclusively for the sampling sites whose microbiome composition was significantly different to South Africa in the pairwise comparisons, we performed differential abundance analyses on the most abundant ASVs to identify the differences at the genus taxonomic level. The comparisons between South Africa and Spain did not detect any significant changes with the threshold implemented (FDR < 0.01). Indeed, the genus Klebsiella  was the only altered in high abundance between South African and Colombian samples, while 48 bacterial ASVs were evidently different in the analysis between Brazil and South Africa sampling sites (Fig. 8). Worth mentioning is that samples from both localities were collected in guava and presented a similar relative abundance in the genus Flavobacterium.  The genus Gluconobacter  and unclassified Halomonadaceae were detected only at the low relative abundance in all medfly samples collected in South Africa. In contrast, a high abundance of genus Acinetobacter , unclassified Burkholderiaceae, unclassified Clostridiales, Dysgonomonas  and Escherichia-Shigella,  was harboured exclusively across all the medfly samples collected in Brazil (Fig. 8).